View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by DigitalCommons@University of Nebraska

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

USDA National Wildlife Research Center - Staff U.S. Department of Agriculture: Animal and Publications Plant Health Inspection Service

2010

Evaluation of harassment of migrating double-crested cormorants to limit depredation on selected sport fisheries in

Brian S. Dorr USDA/APHIS/WS National Wildlife Research Center, [email protected]

Ashley Moerke

Michael Bur

Chuck Bassett

Tony Aderman

See next page for additional authors

Follow this and additional works at: https://digitalcommons.unl.edu/icwdm_usdanwrc

Part of the Environmental Sciences Commons

Dorr, Brian S.; Moerke, Ashley; Bur, Michael; Bassett, Chuck; Aderman, Tony; Traynor, Dan; Singleton, Russell D.; Butchko, Peter H.; and Taylor, Jimmy D. II, "Evaluation of harassment of migrating double- crested cormorants to limit depredation on selected sport fisheries in Michigan" (2010). USDA National Wildlife Research Center - Staff Publications. 904. https://digitalcommons.unl.edu/icwdm_usdanwrc/904

This Article is brought to you for free and open access by the U.S. Department of Agriculture: Animal and Plant Health Inspection Service at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in USDA National Wildlife Research Center - Staff Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. Authors Brian S. Dorr, Ashley Moerke, Michael Bur, Chuck Bassett, Tony Aderman, Dan Traynor, Russell D. Singleton, Peter H. Butchko, and Jimmy D. Taylor II

This article is available at DigitalCommons@University of Nebraska - Lincoln: https://digitalcommons.unl.edu/ icwdm_usdanwrc/904 Journal of Great Lal(es Research 3fi (2010) 215-223

Contents l~stsavailable at Sc~enceD~rect Journal of Research

All$ )_ journal homepage: www.elsev~er.corn/locateljglr

Evaluation of harassment of migrating double-crested cormorants to limit depredation on selected sport fisheries in Michigan

S. Dorr Brian Ashley Moerlte ', Michael Bur ', Chuclt Bassett ", Tony Aclerman e, Dan Traynor b, Russell D. Singleton a, Peter H. Butchlto ', Jimmy D. Tayior I1 " U.S. Department of Agriculture, Wildlife Services. National Wildlife Research Center, P.O. Box 6099. Mississippi State University, A4S 339762, USA State Universio~.650 W, Eostcrdoy Ave.. Sault Ste. Marie. MI 49783. USA USGS Great Lakes Science Center; Lal(e Erie Biological Station. 6100 Colun~busAve. Sar~duslp.OH 44870-8329. USA 9.S. Forest Service. 2727 N, Lincoln Road, Escanaba. MI 49829. USA ' Wildlifc Services, 1865 ORourke Blvd., Suite C, Gaylord, MI, 49735, USA ' Wildlfe Services. 2803 Jolly Road. Suite 100, Okelnos. MI. 48864. USA "5. Department of wiculture. Wildlife Services, Nat!onal Wildlife Researcl~Center; 9730.6, Lathrop Industrial Drive SW, Olympia. WA 98512. USA

ARTICLE INFO ABSTRACT

Article hbtory: Diverse management techniques have been used to mitigate conflicts between humans and double-crested Received 3 June 2009 cormorants (Phalacrocorau auritus) including harassment methods supplemented by lethal take. In this study Accepted 17 November 2009 we evaluated impacts or programs to harass spring migrating cormorants on the (Sander vitreus) fishery. in Brevoort Lake and the yellow perch (Percajlovescen~)and walleye fisheries at Drummond Island. Communicated by Martin Stapanian Cormorant foraging declined significantly (pc0.05) at both locations subsequent to initiation of harassment index words: programs. Overall harassment deteired.90% of cormorant foraging attempts while taking less than 64: Cormorants lethally on average at each site. Yellow perch were a predominate prey item in number and biomass at both Control locations. Walleye made up a small proportion of the diet at both locations. However, both walleye and Management yellow perch abundance increased significantly (pc0.05) at Drummond Island. Walleye abundance at age 3 Migration increased to record levels in 2008 following 3 years of cormorant management at Brevoort Lalce. The Predation estimated cormorant consumption of age-1 walleye in the absence of management at Brevoort Lake during Fisheries 2005 would account .for 55% of the record 2006 age-1 walleye population. These results support the hypothesis,that cormorant predation on spawning aggregations of sporffish was a significant mortality factor and cormorant management reduced sportfish mortality and increased abundance at both locations. Continuation of harassment programs and .fishery assessments will determine whether improvement of targeted sport fisheries through control of spring migrating cormorants is sustainable. Published by Elsevier B.V.

consumption of age 1-2 yellow perch (PercaJavescens) could reduce future angler harvest of yellow perch and to a lesser extent, The interior population of double-crested cormorant (Pl~alacro- (Sander vitreus). R~~dstamet al. (2004) indicated that cormorants corm nulitus; hereafter cormorant) has shown a substantial resur- were a major factor contributing to the decline in walleye and yellow gence over the past35 years (Wires and Cuthbert, 2006). Cormorants perch populations in Oneida Lalte, New York. Fielder (2008) have increased from approximately 32.000 breeding pairs in the mid- concluded that cormorants were an influential factor affecting a 1970s to more than 226,000 breeding pairs (includingthe Great Laltes collapse in the yellow perch sport fishery in the Les Cheneaux Islands, states and provinces) by the mid-1990s (Wires ahd Cuthbert, 2006). , Michigan. Research also indicates that much of this The increase in the interior population of cormorants has caused predation on yellow perch and walleye occurs in the spring. perceived and ltnown impacts to both commercial and natural presumably when these species are aggregated during spawning resources (Taylor and Dorr, 2003). and potentially more vulnerable to predation (Diana et al., 2006; Cormorants have caused documented impacts locally to recrea- Relder, 2008; Rudstam et al., 2004). tional fisheries (Fielder, 2008; Johnson and Raloczy, 2004; Rudstam Drummond Island and Brevoort Lalte, Michigan are locations where et al., 2004). VanDevalli et al. (2002) found that cormorant spring predation by cormorants on spawning yellow perch and walleye was considered to potentially impact those sport fisheries. Fish community assessments have been made in St. Marys River (SMR) ': Corresponding author. from 1975 to 2007, which includes Potagannissing Bay adjacent to E-mail address: brian.s.dorrQaphis.usda.~ov(S.D. Brian). Drurnrnond Island, Michigan (Fielder et al., 2007). In a 2002 study of

0380-133015 - see front matter. Published by Elsevier B.V. doi:lO.1016~jjg1~.2010.02.001 216 S.D. Brian et al. /Journal ojGreat ILnkes Research 36 (2010) 215-223 population characteristics within the SMR, total annual mortality for hypothesis was that predation by spring migrating cormorants was a yellow perch increased and growth rate improved for both yellow perch Iimitingmortality factor on walleye and yeIIow perch at these locations. and walleye in the Potagannissing Bay reach of the SMR (Fielder et al., The objective of this study was to evaluate the effectiveness of 2003). Concomitant with the increase in yellow perch mortality and harassment program with respect to reducing foraging by migrating increased growth rate was the concern of increased predation by cormorants and to evaluate fishery response using data from ongoing cormorants as one of several factors potentially affecting the fish fishery and fish population monitoring programs. population and fishery (Fielder et al.. 2003). At Brevoort Lake walleye abundance, survival, and recruitment Methods have been monitored regularly by the U.S. Forest Service (USFS) and Michigan Department of Natural Resources (MDNR) since the early Study area 1980s. During 1980-1983 the walleye fishery in Brevoort Lake was maintained by stoclting fry and fingerlings as there was little natural Dmmmond Island is located between the Upper Peninsula of reproduction (MDNR unpublished data). Average survival of stoclted Michigan, Chippewa County, and , (Fig. 1). The SMR fingerlings to spawning ages 3 and 4 was 5-28% resulting in a sustainable sport fishery (Bassett, 2006). A spawning reef was constructed in 1984 and a 1989 fishery assessment found that 49%of the adult walleye population in the lake originated from natural reproduction (Base& 2006). Due to these findings, walleye stoclting was discontinued from 1990 to 1996. Subsequent assessments indicated adequate reproduction of walleye but unlilte the pre-reef period, survival to spawning ages 3 and 4 was poor (1-3%: Bassett, 2006). Consequently, numbers of adult walleye declined steadily after 1991. even after walleye stodung was resumed in 1997. Fishery assessments from 1994 to 2005 indicated high mortality of walleye was occurring between fall age-0 and spring age-3 corresponding to total lengths of 13-36 cm. Estimated numbers of spring age-3 walleye declined from a high of 3084 in 1986 to a low of 52 in 2005 (C. Bassett USFS, unpublished data). This decline occurred despite substantial natural reproduction and stoclcing of age-0 walleye between 1991 and 2003 (C. Bassett USFS, unpublished data). Walleye in the affected size range could not be legally harvested by anglers (38 cm minimum length limit) and there were no apparent changes in habitat characteristics or fish community composition that explained higher mortality of young walleye (C. Bassett USFS, unpublished data). However, daily cormorant sightings on the lake increased from less than 20 during the mid 1980s to several hundred by the mid-1990s (C. Bassett USFS, unpublished data). Due to the aforementioned declines in sport fisheries at these locations spring cormorant harassment programs were initiated by the U.S. Depart- ment of Agriculture, Michigan Wildlife Services (WS) at Drummond Island in 2004 and Brevoort Lake in 2005. Fear-provolang harassment methods are non-lethal management techniques used when target speaes are mobile and damage occurs over a limited time period (ep spawning fish and migrating buds). Fear- provolung stimuli (e.g., propane cannons, Mylar tape, human effigies) can protect small areas, although their effects usually are temporary (Conover, 2002). Fear-provoking harassment methods are sometimes integrated with limited lethal take to prevent habituation to harassment techniques (Conover, 2002). Harassmenttechniques such as boat chases and pyrotechnics have been successfully implemented to protect local fisherv resources in New York (Chiuman et al.. 2000). However these techniques when applied to m~lti~leiocationscan be logisticallydifficult, labor intensive, and expensive to imulement effectivelv. In order to address the logistic; and cost, the u.;. Department of Agliculture, Michigan Wildlife Services (WS) implemented a variation of harassment techniques in which volunteer designated agents - Chrrstensen Bay (DAs), are enlisted to protect fishery resources under the provision of I N the US Fish and Wildlife Services Public Resource Depredation Order A (USFWS, 2003). This cooperative effort allows WS to leverage their staff and financing and expand their operations to multiple areas experiencing conflict with migrating cormorants. The harassment programs at Dnlmmond Island and Brevoort Lake were designed to protect spawning assemblages of walleye and yellow Fig 1. a) Locations of Brevaort Lake and Dmmrnond Island, MI. b) Stom Bay and Maxton Bay perch from cormorant predation during spring migration. Harassment sprin~harassment sites of double-crested cannoma evaluated in 2004-2007, Drummond programs included the use of non-lethal techniques (e.g. pyrotechnics, Island, MI. c) Brevmrt Lake. MI, artificial spawning reef, Baedne Bay, and Chrisrensen Bay boat chases) reinforced with limited lethal talte. The underlying spring harassment sires of doublwested cormorants evaluated in 2005-2007. MI. ID. Brian el al. ;journal oJGreat Lakes Research 36 (2010) 215-223 217' flowing down from Lake Superior divides with water flowing South Movements of birds can cause instantaneous counts to underes- into Lake Huron and East to the . The mouth of the timate actual numbers of individuals in a given area (Granholm, '1983; I'otagannissing River, Scott Bay and Maxton Bay were used for spring Lehnen and Icrementz, 2005). Therefore, peak instantaneous counts harassment and food habits collection sites on Drummoncl Island. The were adjusted upward by the average daily turnover rate (0.282) at area is shallow with a n~aximun~depth of about 4 rn and a total area of stopoversites of cormorants duringspring migration based on satellite about342 ha. Brevoort Lalte has asurface area of 1712 ha and is located telemetry data (data source: USDA, Wildlife Services. National Wildlife on the USFS, Hiawatha National Forest, Macltinac County, Michigan Research Center. J.D.Taylor TI). Stopciver duration is the length of time (Fig. 1 ). Most of the shoreline is forested and shallow areas and shoals that a migrating bird remains at a stopover location during migration are primarily sandy grading to pulpy peat in deeper water. The lake is (Piersma, 1987). The total number of individual cormorants using a relatively shallow with a maximum depth of about 10 m. Both Brevoort spring harassment location was calculated as: Lalte and Drummoncl Island have a variety of fish species with the primary sportfish species being walleye, yellow perch, TN = C(IDC* (1 + ADTR)) (Box lucius), and smallmouth bass (Micropterus dolomieu). where: TN=total number of individual cormorants during the harassment period, IDC=pealc instantaneous daily count, and Harassment evaluation ADTR= average daily turnover rate at spring migration stopovers. The total number of cormorants represents the total number of Harassment was initiated In 2004 at Drummond Island and2005 at cormorant days of predation that could have occurred in the absence Brevoort Lalte and occurred each year between April 9 and May 13, of harassment. Therefore fish consumption estimates are a projection from dawn to dusk. Participants maintained a log ol the number of orthe maximum that could have been consumed based on observed cormorants observed, start and end times, number of cormorants counts and diet data. The net number deterred from foraging Itilled, number recovered, number of shot shells used, and number of represents the effective harassment effort in terms of reduction in pyrotechnics used. A harassment period is defined as the time an cormorant days of foraging. An ANOVA with Tultey's Studentized individual(s) started active harassment to the time they were either range (HSD) test was used to evaluate the differences in mean daily replaced by another individual(s) or active harassment ended (e.g. all instantaneous cormorant counts between years for each location cormorants left or dusk). Total hours of harassment represent the (Proc GLM. SAS, 1994). total person hours (i.e. 2 people at 4 h each =S k). Harassment intensity is reported as the average hours of active harassment effort Consumption estimates per hectare per day (h/ha/d) at each site for all years. Wildlife Services staff or DAs estimated the numbers of cormorants in each Food habits flock dispersed, and the number that were not successfully dispersed Stomachs were collected from up to 10 cormorants each day of (i.e. were able to forage) during each period. Estimates for each period spring harassment. Cormorants were shot with 12 gauge shotguns were summed to provide the total number of foraging attempts made using non-toxic shot. Immediately after collection stomachs were by cormorants at a given site and date. Similarly, estimates for each preserved by injecting 60 ml of formalin and stored in labeled bags on period were summed to provide the total number of non-deterred ice until samples could be stored in a freezer. foraging attempts. The net number of foraging attempts deterred was Procedures described by Bur el al. (1999) were followed lor food used as a measure of the effective harassment effort in terms of per- habit analysis. Contents of each stomach were removed and weighed. centage reduction in foraging attempts. Each prey item was identified from whole, partial (e.g. bacltbone with flesh), and fragments (otoliths and diagnostic bones) of specimens. Estimation of cormorant foraging Total length of prey items was measured to the nearest millimeter. Standard and bacltbone lengths from partially digested fish were ~~~i~~~i~~the total number of foraging attempts at a spring converted to total lengths from standard equations and fish weights harassment site lilcelll overestimates the number of individual were determined from length weight regressions (BaYley and Austen, cormorants attempting to forage at a given site. Tllis is due to 1987; Buret al., 1999; Cameronet al., 1973: Carlander, 1969; Knight et fact that an individual cormorant may repeatedly try to forage in a d.,1984; ODNR, 1997,2005: Schneider et al., 2000; WesleY, 1996, and given harassment location. To address this issue of overestimation Great Science center, "Ite unpublished data). peak instantaneous counts adjusted for turnover rate were used to Estimates number and biomass prey each year provide an estimate of the number of individual cormorants using a were based On: site (Lehnen and Icrementz, 2005). To avoid double counting cormorants, instantaneous counts were used from one observer per location per period. The observer scanned where T= total number of prey i consumed per harassment period j. the harassment site at the beginning ol each harassment period and and d=number of days during the harassment period j, ni=mean recorded the number of cormorant observed at that time. Only cor- number of prey i per harassment period j, and N= total number of morants within the harassment area whether on the water or flying cormorants per harassment period j; were counted. Maxton Bay and Scotts Bay were the l~arassmentsites at Drummond Island (Fig. 1) and could be scanned from a single location by land or boat, Cormorants at Brevoort Lalce were counted by driving a boat across the lake to observe the entire area. The focus of where F=number of daily feedings per harassment period j, harassment activities were Boedne Bay, Christensen Bay, and the gi= mean weight of stomach contents per harassment period j, and artificial spawning reel (Fig. 1) although cormorants were harassed 0.55 kg= mean daily food consumption per cormorant (Seefelt, over the extent of the lake. The maximum instantaneous count of 2005); and cormorants deterred during the harassment period from any one observer was used as the minimum number of cormorants present at the site. The peak instantaneous count of cormorants reported as not deterred from foraging at the end of each harassment period from any where B=biornass of prey i during harassment period j and one observer was used as the peak total not deterred. wi = mean weight of prey i during harassment period j. 218 59. Brian et a1 /Journal oJGreot Lakes Research 36 (2010) 215-223

Daily food consumption (DFC) of 0.55 Icglcormorant was talten Table 2 from Seefelt (2005) for adult cormorants during the pre-nesting/ Location, dates of harassment, date of peak estimated number of double-crested cormorants observed, total estimated number of cormorants observed, and total incubation period. In the absence of DFC data specific to spring estimated deterred from foraging over the harassment period, at Brevoort Lalte, MI. migrating cormorants the pre-nestinglincubation period data were 2005-2007 and Drummond Island. MI. 2004-2007. Numbers of individual cormorants used to best represent energetic needs. These consumption estimates were estimated from peak instanraneous daily counts ad,iusted for average daily differed from Buret al. (1999) in that all cormorants were considered turnover rate. to be adult migrating cormorants. Location Dates of Daily peak Total number Total number of harassment count date of cormorants cormorants Drummond Island fishery assessment deterred from foraging Data from the 2002 and 2006 MDNR SMR fishery assessments Brevoort 4/16 to 5112,2005 4/24/2005 18.495 17.855 were used to evaluate fishery response to the cormorant harassment Lake Brevoort 4/13 to 5/13,2006 4/25/2006 2349 1709 program at Drummond Island. Comparisons between assessment Lake periods included only the Potagannissing Bay data from the SMR Brevoort 4/17 to 5/11,2007 4/28/2007 3179 2376 because the mouth of the Potagannissing River, Maxton Bay, and Lake Scotts Bay are important spawning locations for walleye and yellow Drummond 4/22 to 5/13.2004 4/28/2004 36.409 33.582 perch and are within the Potagannissing Bay assessment area. The lsland Drummond 4/13 to 5/12.2005 4/23/2005 9446 7687 2002 and 2006 assessments were used because they had the same lsland gill-net mesh sizes so catch-per-unit-of-effort (CPUE) was fully Dn~mmond 419 to 5/12.2006 5/3/2006 5200 2928 comparable between years. The sample size was 10 net sets at the lsland same locations each year. A t-test was used to determine significant Drummond 4/20 to5/12,2007 4/27/2007 13346 10,147 (~~0.05)differences in CPUE between assessment years. We lsland conducted Levene's test for equality of variances and report approximate t values using individual sample variance and Sat- terthwaite's approximate df values in cases of unequal variance sampled ranged from 5.5 to 8.4 Icm. Abundance of age-0 and age-1 (Norusis, 1990). Total annual mortality derived from catch curve walleye captured by electrofishing was estimated by the Serns analyses was qualitatively compared between assessment years for method (Serns, 1982, 1983). Survival-of age-0 and age-1 walleye yellow perch. was determined from the Sems' estimates, Michigan DNR stoclung records and subsequent Schnabel estimates of adult abundance. Brevoort Lake walleye assessment Results Data from spring trap-netting during 1985-2008 were used to assess effects of cormorant harassment on abundance of adultwalleye Drummond Island harassment evaluation (ages 3 and older). Shortly after ice-out, 5-6 nets were set on the Bladc Point spawning reef and one net was set in Christensen's Bay or the There was a total of 989 h (R = 247/period, SD = 103) of mouth of the Little Brevoort River. The reef is the primary spawning harassment between April 9 and May 13, for 2004-2007 (Table 1). site in the lake but pre-spawn congregations of walleye occasionally Averaged over all year mean intensity of effort was 0.03 h/ha/d occur at the other sites. Nets were inspected daily except'when high (SD=0.01). The total number of shotgun shells used was 3495 winds prevented access. Walleye captured for the first time were (R=874/period, SD=439), the total pyrotechnics used was 1558 marked with an upper caudal fin clip. All walleye were transported (%= 390/period, SD = 378), and total number of cormorants ItiIled about 0.8 Im from the net site for processing and release. Abundance was 719 (8=180/period, SD=89: Table 1). The total number of was estimated by the Schnabel multiple census method (Riclter, connorant foraging attempts estimated for 2004-2007 was 105,905 1975). Age and growth analysis from scale samples provided the basis (R = 26,476/period, SD =28,850). The total net foraging attempts for walleye population estimates by age groups. Beyond age 6, two or deterred were 95.653 (X =23.913/period, SD = 28,252) with on more age groups were often combined due to the difficulty of average 90%of the total foraging attempts deterred (Table 1). Mean accurately aging scales from older walleye. instantaneous counts per harassment period between April 9 and May Fall electrofishing data collected during 2006-2008 and periodi- 13, from 2004 to 2007 was 312 cormorants (SD=473), cally through the 1990s provided additional information on abun- Based on instantaneous daily counts and daily turnover rate, a total dance and survival of juvenile walleye. Electrofishing was conducted of 64,401 (R= 16,10O/period, SD= 13,912) individual cormorants during early October with a Smith-Root SR-16 unit operating at 10- were estimated to have attempted to forage from 2004 to 2007 12 A pulsed DC. Sampling occurred in depths of 0.61-2.13 m at water (Table 2). Of the estimated total a net of 54,344 (R= 13,586/period, temperatures ranging from about 10 to 15.6 "C. Length of shoreline SD= 13,663) individual cormorants were prevented from foraging by

Table 1 Total number of double-crested cormorant foraging attempts (i.e. individual cormorants may have been harassed >1 time), total foraging attempts prevented, harassment effort. number of shotgun shells and pyrotechnics used. and number of cormorants taken lerhally from Brevoort Lake, MI, 2005-2007 and Drummond Island, MI. 2004-2007. Location Dates of Total foraging Total foraging attempts Harassment Number of shells Number of harassment attempts prevented (%) effort (h) used (pyrotechnics) cormorants taken Brevoort Lake 4/16 to 5/12,2005 22.286 21.621 (97) 1784 1854 (1400) 637 Brevoort Lake 4/73 to 5/13.2006 8218 6435 (78) 7181 901 (839) 271 Brevoort Lalte 4/17 to 5/11.2007 9924 8074 (81) 1153 1503 (746) 380 Drummond Island 4/22 to 5/13.2004 69.124 65.722 (95) 311 1374 (661) 293 Drummond Island 4/13 to 5/12.2005 14,022 12.339 (88) 352 924 (769) 188 Drummond Island 419 to 5/12.2006 5584 3527 (63) 200 893 (54) 162 Drummond Island 4/20 to 5/12.2007 17.175 14.065 (82) 126 304 (74) 76 S.D. Brian et al. i Journal of Creot Lakes Research 36 (2070) 275-2173 21 9

2,5001 Brevoort Lake harassment evaluation

There was a total or41 18 h (8= 1373/period, SD = 357lperiod) of harassment from April 16 to May 13, from 2005 to 2007 (Table 1). Averaged over all years mean intensity of effort was 0.03 h/ha/d (SD=O.Ol). The total number of shotgun shells used was 4258 (R = 1419/period, SD = 482/period), the total pyrotechnics used was. 2985 (2=995/period, SD=354/period), and total number of cormorants ltilled was 1288 (R =429/period, SD = 18S/period; Table 1).The total number of cormorant foraging attempts estimated for the period 2005-2007 was 40,428 (2= 13,47G/period, SD = 76771 period). The total net foraging attempts deterred were 36,130 (8=2043/period, SD=8335/period), with on average 89% of total foraging attempts deterred (Table 1 ). Mean instantaneous counts per harassment period between April 16 and May 13, from 2005 to 2007 was 75 cormorants (SD=.l58). Based on instantaneous daily counts and daily turnover rate a total of 24,023 (8= 800S'/period, SD = 9092lperiod) individual cormorants were estimated to have attempted to forage for 2005-2007 (Table 2). A net of 21,970 (8= 7323/periocl, SD= 9153lperiod) individual cormorants were prevented from. foraging by harassment efforts. 2005 2006 2007 Peak foraging attempts for 2005-2007 occurred between April 24th Year and April 28th (Table 2). Mean daily counts of foraging cormorants declined significantly (2-way ANOVA, F2. 83 =29.06, p=<0.0001) in Fig. 2. a) Mean daily counts (bars) of double-crested cormorants using the Drummond years subsequent to the initiation of harassment (Fig. 2). On average Island area of Lalte Huron. MI, during spring miiration 2004-2007. b) Mean daily about 5.45; of the total estimated individual number of cormorants counts (bars) of double-crested cormorants using Brevoort Lake, MI, during spring migration 2005-2007. Vertical lines represent 95%confidence interval estimates. Bars were talten lethally during the harassment programs: with different letters are significantly different (p.cO.05) from each other. Food habirs Drummond island

The cormorant diet near Drummond Island during 2005 and 2006 harassment efforts (Table 2). Peak foraging attempts for 2004-2007 (April-May) consisted of 13 species and two taxonomic groups occurred between April 23rd, and May 3rd (Table 2). Mean daily Coregonidae (generally whitefish Coregonus clupeafomis) and Catos- counts of cormorants declined significantly (2-way ANOVA, F3, = tomidae (mostly white suclter Catostomus commersoni) (Table 3). 43.51, p = <0.0001) in years subsequent to initiation of harassment Cormorant diets were made up primarily of yellow perch during both (Fig. 2). On average about 1.1% of the total estimated individual years. Because diet data were not collected in 2004 and 2007 the number of cormorants were taken lethally (including food habits average observed diet from collections conducted in 2005-2006 were collections) during the harassment programs. used to estimate cormorant prey consumption for 2004 and 2007.

Table 3' Projected annual number and biomass of prey thatwould have been consumed in the absence ofharassment by all cormorants observed compared to actual observed consumption of foraging cormorants (in parentheses) in Drummond Island. Lake Huron. MI. 2004-2007. Only taxa with --I%relative abundance shown (except for walleye). Unidentified species also not included. Species (common name) Number (thousands)/year Biomass (1tg)lyear 2004' 2005 2006 2007" 2004" 2005 2006 2007' Coregonidae (whitefishes) 4.3 2.7 0.0 1.7 645.0 400.9 0.0 2483 (0.3) (0.4) (0.0) (0.3) (39.1) (58.2) (0.0) (46.4) Box lucius (northern pilce) 4.3 1.8 0A 1.7 658.7 163.8 98.5 253.5 (0.3) (0.3) (0.1 ) (0.3) (39.9) (23.8) (33.6) (47.4) Notropis hudsonius (spottail shiner) 17.2 5 3 2.1 6.6 84.6 23.2 11.8 32.6 (.lo) (0.8) (0.7) (1.2) (5.1 (3.4) (4.0) (6.1) Cyprinidae (minnows/carp) 144.8 49.0 16.4 , 55.7 1035.0 296.7 138.6 ' 398.4 (8.8) (7.1) (5.6) (10.4) (62.7) (43.1 ) (47.2) (74.5) Catostomidae (suclters) 127.6 16.9 25.0 . 49.1 8759.4 1590.7 15292 3371.3 (7.7) (2.5) (8.5) (9.2) (530.5) (231.0) (521.1 ) (630.3) Ameium nebuiosw (brown bullhead) 1.4 0.9. 0.0 0.6 228.1 141.8 0.0 87.8 (0.1) (0.7 (0.0) (0.1) (13.8) (20.6) (0.0) (16.4) Centrarchidae (sunfishes) 31.5 18.7 0.4 12.1 204.1 121.8 2.0 78.6 (1.9) (2.7) (0.7) (2.3) (12.4) (17.7). (0.7) . (14.7) Percaj7avescens (yellow perch) 411.4 1203 - 54.3 1583 9448.2 1941.O 1488.3 3636.4 (24.9) (17.5) (18.5) (29.6) (572.2) (281 9) (507.2) (679.8) Sander ~dtreus(walleye) 2.9 1.5 0.0 1.1 1303 81.0 . 0.0 50.2 (0.2) (0.3) (0.0) (0.2) (7.9) (11.8) (0.0) . (9.4) Other spp. 11.5 4.5 1.1 4.4 37.0 9.9 53 14.2

(0.7) (0.4) (0.8) (2.2) . (1.4) ' (1.8) (2.7) Total 756.9 221.9(0.6) 99.7 291.3 21.230.4 4770.6 3273.7 8171.0 [45.8) (32.2) (34.0) (54.5) (7285.8) (692.9) (1115.6) (1527.6) " Diet data were not collected in 2004 and 2007. The average cormorant diet composition for 2005-2006 was used for 2004 and 2007 estimates. 220 S.D. Brian et a1 1)oumal of Great Lakes Research 36 (2010) 215-223

Projected numerical consumption of all fish species by cormorants in the absence of harassment ranged from 99.7 to 756.9 thousandlyear and was 3 to S fold higher in 2004 than in 2005-2007 (Table 3). Numerically yellow perch, Cyprinidae, and Catostomidae were the most plentiful diet items. Projected biomass consumed by cormorants ranged from 3273.7 to 21.230.4 kg/year and was 3 to 6-fold higher in 2004 compared to 2005-2007 (Table 3). The projected biomass of prey consumed followed a similar trend as abundance with yellow perch, Catostomidae, and Cyprinidae comprising the greatest weight of prey consumed. The projected biomass of yellow perch consumed Year in 2004 was 9448.2 kg and in 2006 was 1488.3 kg. The projected Fig. 3. Fa11 electrolishing record of numbers or age-1 walleye in Brevoort Lake, MI. biomass of walleye consumed in 2004 was 130.3 kg. No walleye were 1991-2008. found in the diet in 2006. Numerical consumption by foraging cormorants only was on average 8-fold less than projected foraging activity by all cormorants in the absence of harassment activities comprising the greatest biomass (49-66%). Northern pilce was the (Table 3). Estimated biomass consumed by foraging cormorants only next largest contributor to projected biomass in 2005, whereas round was on average 8-fold less than projected foraging by all cormorants goby (Neogobius melanostomus) and Centrarchidae were in 2006 and in the absence of harassment activities (Table 3). 2007, respectively. Numerical consumption by foraging cormorants only was over all years 19-fold less than the projected foraging activity by all cormorants in the absence of harassment activities Food habits Brevoort Lake (Table 4). Estimated biomass consumed by foraging cormorants only was reduced by 39-fold in 2005 and about 5-fold in 2006 and 2007 The cormorant diet at Brevoort Lake during 2005-2007 included (Table 4). Mean total length of yellow perch consumed by cormorants more than 20 fish species and 1 crustacean. Brevoort Lake cormorants was 78,71, and 91 mm in 2005,2006, and 2007, respectively. Mean consumed an average of 8.6, 18.0, and 12.1 prey items daily per total length of walleye consumed by comorants was 106, 139. and individual collected in 2005-2007, respectively. Projected numerical 309 mm in 2005 through 2007, respectively. consumption of all fish species by cormorants ranged from 108.4 to 7472.0 thousandtyear and was 7 to 14-fold higher in 2005 than in 2006 and 2007 (Table 4). Yellow perch were the most abundant prey Drummond island fishey consumed by comorants and comprised more than 67%of prey items in all years. Crayfish (Astacidae) also were numerically abundant in all Mean survey net CPUE for yellow perch was significantly greater years (25%of diet) but comprised more than 17%of the diet in 2005. (t9,= -2.2, p = 0.04; tdr adjusted for unequal variances) in 2006 Walleye comprised less than 1%of prey items by number for all years. (f=62.9, SE=23.5) than 2002 (f =7.5, SE=3.9). Mean survey net Projected biomass consumed by cormorants in the absence of CPUE for walleye was significantly greater (t18= -2.5, p =0.02) in harassment ranged from 1311.6 to 10.639.3 lg/year and was 6 to 8- 2006 (R=6.5, SE= 1.6) than 2002 (f= 1.8, SE= 1.0). Catch curve fold higher in 2005 compared to 2006 and 2007 (Table 4). Projected analyses of mortality for yellow perch increased from 0.57 to 0.96 biomass trends were similar to numerical trends, with yellow perch from 2002 to 2006. There were not enough walleye sampled across age classes to produce mortality estimates using catch curve analyses.

Table 4 Brevoort Lake fishery Projected annual number and biomass of prey that would have been consumed in the absence of harassment by all cormorants observed compared to actual observed Fall 2006 electrofishing walleye catch was 3.7 times that of the consumption of foraging cormorants (in parentheses) in Brevoort Lake. MI 2005-2007. previous high catch rate (19.25Jlcm vs 5.27/1m) dating back to 1991 Only taxa with >I%relative abundance are shown (except for walleye). Unidentified species also not included. (Fig. 3). Survival of the 2003 walleye year-class to age-5 was 5 fold higher post management than survival of the 2000 walleye year-class Species (common name) Number (thousands)/ Biomass (&)/year to age-5 (Table 5). Spring 2008 netting (N=24) yielded the second year highest adult (age3 and older) walleye population estimate 7780 2005 2006 2007 2005 2006 2007 (95%Cl= 6633-9413) recorded for this lake (Fig. 4). Ninety percent of Astandoe (crayfish) 2603 9.4 52 624.8 22.5 12.5 the adult population consisted of the 2005 year-class which was the (6.7) (2.0) (1.0) (16.1) (4.8) (2.5) first year-class fully recruited Following initiation of management. Esox lucius (northern pike) 93 0 0.2 10753 0 183 (0.2) (0) (~0.1) (27.7) (0) (3.6) Neo~obiusmelanostomus 21.7 11.5 1.2 356.0 1833 19.3 (round goby) Percopsis orniscornaycus Table 5 (trout-perch) Estimated cohort survival of stocked walleye fingerlings in Brevoort Lalte, MI. Coltus bairdii Year-class Number Year of Age Number of adults (mottled sculpin) stocked evaluation Centrarchidae (sunfishes) Total Population Percent catch estimate (95%C.I.) survivala Percapavescens (yellow perch) 1997 20,534 2001 4 78 90 (30-94) 0.4 Sander vitreus (walleye) 2000 22,665 2004 4 9 44 (14-46) 0.2 2005 5 10 16(10-41) 0.1 Other spp. 2003 19433 2008 5 44 90 (56-2351 0.5

a Percent survival in all years may be overestimated because adult population Total estimates include naturally reproduced fish which could not be distinguished from stocked fish. S.D. Brio11 et 01. I ]ourilol ot Greol: Lolcs Research 316 (2070) 2151223 221

inclividual cormorants using each,site, the estimated kg of fish consuniecl declined substantially from 2004 levels at Drummond Island and 2005 at Brevoort hlte.

Food 11abit.s and ,fishery response

Drurnrnond ls[and Comparisons of survey net CPUE of yellow perch and walleye in Potagannissing Bay indicate significantly greater abundance of both species in 2006 (post-cormorant management) than 2002 (pre- Year cormorant management). Mortality estiinates for walleye specific to fig. 4. Spring population estimates and 95%confidence intervals (vertical lines) for aged Potagannissing Bay are not available as the sample size was not large and older walleye in Brevoort Lake 1985-2008. Double-crested cormorant management enough to develop a bay specific mortality estimate. Mortality esti- was initiated in spring 2005. mates from catch. curve analysis indicate mortality of yellow perch increased over the same period. However this increase could be an artifact of unequal recruitment because of the strong 2003 year-class Discussion entering the age structure (Fielder et al., 2007). The violation of the constant recruitment assumption seems liltely given that the yellow Harassment perch mortality estimate (96%)would not be sustainable and should result in a decline in yellow perch in Potagannissing Bay. Effectiveness of cormorant harassment efforts varied between Diet data indicate cormorants consumed 1.36 times the biomass sites and years ranging from 60%to 97%of foraging attempts deterred and ,9 times the number of yellow perch harvested in the.Potagan- within a season. However, on average over both sites, 90% of nissing Bay open water fishery of 1999 (Fielder et al., 2002). While the cormorant foraging attempts were deterred due to harassment fishery and cormorant consumption data are not .directly compar- activities. In addition there was a significant (pc0.05) 79% decline able due to differences in age classes predated or harvested and the in the average estimated number of cormorants 'over both sites for absence of yellow perch population data, they suggest a potential re- years subsequent to initiation of harassment programs. Average cruitment issue for the fishery.This is supported by research showing intensity of effort was similar at both sites at 0.03 h/ha/d of compensatory processes affecting yellow perch survival have occurred harassment effort. Both sites declined in use of shotgun shells, by age-1, and mortality is largely additive on total mortality (Forney, pyrotechnics, and numbers of cormorants talten lethally subsequent 1980; Nielsen, 1980; Itudstam et al., 2004). Additionally, the age-1. to initiation of harassment. The overall decline in numbers may have cohort size is a good predictor of the number of adult yellow perch been affected by control efforts of breeding cormorants in the Les recruiting to the fishery (Forney, 1980; Rudstam et al., 2004). These Cheneaux lslands (LCI), Lalte Huron, ,Michigan and declines in data combined with more detailed demographic data on the perch breeding populations of cormorants in the North Channel of Lake fishery in Potagannissing Bay may clarify possible effects of cormorant Huron, Ontario, CA. Control efforts have been effective in reducing consumption of sub-adult yellow perch. breeding cormorant numbers by more than 73%in the LC1 from 2004 There are a number of factors that can affect the yellow perch and to 2008 and this may affect the use of Brevoort Lalte and Drummond walleye populations other than cormorant predation. Percid repro- Island sites. However, control efforts did not affect the decline in ductive success greatly increased in all of Lalte Huron starting in 2003, observed numbers of cormorants on breeding' colonies as rapidly as presumably due to the collapse of alewives (Fielder et al., 2008). The the decline seen at harassment locations. The more rapid decline at lack of reliable cohort.specific mortality data rnalte the determination harassed sites suggests that harassment alone may have some inter- of whether there are more walleye and yellow perch because of better annual effect on use of the spring stopover sites. This inter-annual reproductive success, lower mortality, or a combination of these decline suggests that cormorants modified migratory behavior to factors difficult (O'Gorman and Burnett, 2001). Given these caveats avoid harassed sites with the effect of malting the programs more trends in yellow perch and walleye abundance are also consistent effective and liltely reducing lethal take of cormorants. Further with the underlying hypotheses that corrnorants are a limiting research would be necessary to evaluate the possibility of modifica- mortality factor and reduction of cormorant caused mortality can tion of migrato~fbehavior due to management specific to migrating increase the abundance of selected'fish stoclts. Future St Marys River corrnorants. Regardless of the causes of the annual decline in number fishery assessments may provide a clearer picture of the effects of of cormorants, the combined non-lethal harassment reinforced by cormorant harassment in Potagannissing Bay. limited lethal harassment was effective in deterring a large percent- age of cormorants at these sites within each year. Brevoort Lake Cormorants displayed little temporal variation in use of each site Estimates of cormorant consumption of prey based on abundance between years. Peak counts of cormorants at both sites occurred and biomass declined from 2005 to 2007 in Brevoort Lalte, mirroring between April 23rd and April 28th of each year, with the exception of declines in cormorant abundance and foraging attempts. Co~morants Drummond Island in 2006, when peak counts occurred on May 3rd. In consumed a diversit)! of prey in Brevoort Lalte which is consistent most cases there wa's a relatively rapid buildup in numbers and a with other food habit studies and supports previous observations that pronounced pealt in numbers of cormorants moving though each area cormorants are opportunistic feeders (e.g. Ludwig et al., 1989). Age-0 and then a more gradual decline in use. This information on temporal and age-1 yellow perch (c90 mm) were the dominant prey item in all pattern could help in refining harassment programs to maximize years in Brevoort Lalte. Declines in total yellow perch consumed effectiveness in limiting cormorant foraging. (abundance and biomass) since the initiation ol control measures Estimates of the individual number of cormorants using Brevoort have been observed. However, the evaluation of the impacts of Lalte and Drummond Island adjusted upward for turnover rate changes in cormorant predation is difficult because very little data on averaged about 62% of the total number of .estimated Foraging the yellow perch fishe~yexist at Brevoolt Lalte. attempts. On average less than 5.4% of the cormorants migrating Walleye comprised numerically less than 1%of the cormorant prey through each site were talten lethally,,based on the estimated number items in 2005-2007, yet this liltely was a significant portion of the of individual cormorants observed. As with the estimated number of existing walleye population in Brevoort Lalce. An estimated 13,900 222 SD. Brian et al. /journal of Great Lakes Research 36 (2010) 215-223

walleye, comprised mostly of age-1 fish, were projected to have been anglers reporting much improved fishing success for yellow perch consumed in the absence of harassment in 2005. Compared to the since 2005 (C. Bassett USFS. unpublished data.). In the long term, a 17 year record age-1 walleye year-class of 2006 (25,480) this level of return to more consistent walleye reproduction at moderate levels cormorant predation woulcl account for 55% of the total age-1 would indicate sustained benefits of cormorant harassment. These population. Cormorants on Brevoort Lake consumed walleye in size results on Brevoort Lake suggest that sustained harassment of classes corresponding to age-1 and age-2. The primarily additive cormorants on inland laltes can improve suppressed walleye fisheries. mortality (Rudstam et al., 2004) and cumulative effects of cormorant Because cormorant harassment programs were conducted concur- predation on a large percentage of age-1 and age-2 walleye in rent with evaluation efforts, by necessity projected cormorant con- multiple year-classes aFe sufficient to account for observed declines in sumption estimates were used.These estirnatesshould be interpreted as recruitment to the spawning population and sport fishery from the the maximum amount of prey that could have been consumed in the mid-1990s to 2005. absence of harassment. Thus, projected consumption at both sites may Cormorant diets indicated reduced individual consumption in overestimate what would happen if management had not been 2005 when cormorant numbers were highest. After control measures implemented. This overestimation could occur for example if cormorant were initiated, the number of prey items consumed by each consumption of a fish species or species cohort reduced the population cornlorant increased as cormorant numbers declined. This increase density to the point where the density was no longer energetically in number of prey items consumed by ind~vidualcormorants may profitable for cormorants to consume them. In the case of reduced prey reflect increased abundance of prey items and consequently increased density cormorantr; would likely switch to other prey or forage foraging efficiency. Reduced intraspecific competition due to man- elsewliere. This change in foraging would be reflected in changes in agement caused declines in cormorant numbers may also have the diet of unharassed cormorants over time as the prey base was increased individual foraging efficiency (Lewis et al., 2001). The size of depleted. Conversely, if the targeted fish speaes were not being yellow perch and walleye consumed also increased over the consumed and observed numbers of cormorants were low then management period. The increased survival of walleye to older age projected fish consumption estimates would be negligible. This result classes and consequently greater abundance of larger fish may have would exonerate cormorants as an influencing factor on targeted fish shifted the mean weight of walleye consumed by cormorants upward. populations. The projected consumption estimates in this study clearly The larger size of individual yellow perch consumed suggests indicate that cormorants are capable of consuming a large proportion of increased survival and abundance of yellow perch although empirical cerfain age classes and fish populations even when the species fishery data are laclting. or age class maltes up a small percentage of the cormorant diet The record 2005 walleye year-class in Brevoort Lalte could not have Non-lethal harassment programs using designated agents and happened unless environmental conditions favored high egg hatching supplemented with limited lethal control have been successful in success and fry survival. However, additional periods favoring high reducing cormorant predation in this study. Harassment resulted in walleye egg hatching success were liltely over the 17year monitoring both reduced predation within years and significant (p<0.05) period, but this success was not reflected in increased recruitment to reductions in predation between years. Evaluation of diet data age3 or older age classes. Cormorant predation on this lalte only occurs indicated that sportfish were being consumed in numbers and for about a month during the spring. Consequently, this predation has biomass that could impact recruitment to the fishery. Fishery data little orno direct influence on walleye survival until the second and third for both sites indicated that targeted fish populations increased in years of life (ages 1 and 2) when walleye are large enough to be of abundance concomitant with the decline in cormorant predation. interest to cormorants relative to other food sources such as yellow Fisheries response in this study is consistent with the underlying perch. The results of this study suggest that sustained cormorant hypothesis that cormorant predation was a significant mortality harassment allowed substantially higher survival of walleye to age3 factor. However, cormorants are only one of many possible factors than would have occurred otherwise. affecting these fisheries. Continuation of harassment programs and Cormorant food habitats (this study) and numbers of cormorants fishery assessments will determine whether improvement of targeted observed on Brevoort LaIte during the harassment effort suggest that soort fisheries through cormorant control is sustainable. cormorant predation is a viable explanation for the sharp decline in ~arassmentprograms are not applicable in all cases. Brevoort Lake walleye survival to spawning age that occurred during the 1990s. The and Drurnmond Island reflect situations were a combination of factors unprecedented high survival of the first walleye year-class (2005) to converge that make harassment a viable management method. These be protected from heavy cormorant predation by a harassment effort factors include vulnerable spawning fish stocks, a relatively limited provides further evidence of a link between cormorant abundance area to be harassed, a large number of migratory fish-eating cor- and walleye survival in Brevoort Lake. morants arriving concurrent with spawning, and a pool of dedicated Record numbers of age3walleye in 2008 may reflectthe dual effects and willing volunteers to undertake the considerable harassment of reduced predation by cormorants on juvenile walleye and reduced effort Cormorant management may not be needed or effective in all competition and predation by other fish species (including adult situations and alternative factors should be considered carefully prior walleye). Ten to 15years of heavy cormorant predation may have to establishing control programs. suppressed the entire fish community in Brevoort Lake, reducing the potential for competitive and predatory interactions between juvenile walleye and other species (Rudstam et al.. 2004). In a sense, a "vacuum" is created in the fish communitythat can be filled by the first strong year- The USDA, Wildlife Services designated agents at Brevoort Lake class of any species that is no longer subject to cormorant predation. and Drummond Island provided considerable time and effort in Additional years of monitoring are needed to determine if the support of this research which are greatly appreciated. Extensive help rebound of the Brevoort Lalte walleye fishery will be sustained. The in the field and with data collection and entry was provided by G. dominant 2005 walleye year-class will exert substantial competitive Rigney, P. Ryan, A. Wilson, J. Hill, T. Harris, S. Lemmons, I<. Hanson, S. and predatory influence on succeeding walleye year-classes for Woodruff, and P. Fioranelli. D. Fielder with Michigan DNR reviewed several years. Electrofishing catch data indicate low walleye repro- this manuscript and assisted with Drummond Island fishery data and duction since 2005 (Fig. 3). Competitive influences of increases in analyses. I<. Hanson, M. Smith, T. IGng, B. Blaclwell, D. Schlosser, and populations of other fish species such as yellow perch also reduce the 3 anonymous reviewers also provided helpful manuscript reviews. probability of having another large walleye year-class during the next This article is Contribution 1581 of the USGS Great Lakes Science few years. An increase in yellow perch is supported anecdotally by Center.